The Astrophysical Journal, 703:L99–L103, 2009 October 1 doi:10.

1088/0004-637X/703/2/L99

C 2009. The American Astronomical Society. All rights reserved. Printed in the U.S.A.

HAT-P-7: A RETROGRADE OR POLAR ORBIT, AND A THIRD BODY

Joshua N. Winn1 , John Asher Johnson2,7 , Simon Albrecht1 , Andrew W. Howard3,4,8 , Geoffrey W. Marcy3 ,
Ian J. Crossfield5 , and Matthew J. Holman6
1 Department of Physics, and Kavli Institute for Astrophysics and Space Research, Massachusetts Institute of Technology, Cambridge, MA 02139, USA
2 Institute for Astronomy, University of Hawaii, Honolulu, HI 96822, USA
3 Department of Astronomy, University of California, Mail Code 3411, Berkeley, CA 94720, USA
4 Space Sciences Laboratory, University of California, Berkeley, CA 94720, USA
5 Department of Physics and Astronomy, University of California, Los Angeles, CA 90095, USA
6 Harvard-Smithsonian Center for Astrophysics, 60 Garden St., Cambridge, MA 02138, USA
Received 2009 August 12; accepted 2009 August 21; published 2009 September 8

ABSTRACT
We show that the exoplanet HAT-P-7b has an extremely tilted orbit, with a true angle of at least 86◦ with respect
to its parent star’s equatorial plane, and a strong possibility of retrograde motion. We also report evidence for
an additional planet or companion star. The evidence for the unparalleled orbit and the third body is based
on precise observations of the star’s apparent radial velocity (RV). The anomalous RV due to rotation (the
Rossiter–McLaughlin effect) was found to be a blueshift during the first half of the transit and a redshift
during the second half, an inversion of the usual pattern, implying that the angle between the sky-projected
orbital and stellar angular momentum vectors is 182.◦ 5 ± 9.◦ 4. The third body is implicated by excess RV
variation of the host star over 2 yr. Some possible explanations for the tilted orbit of HAT-P-7b are a close
encounter with another planet, the Kozai effect, and resonant capture by an inward-migrating outer planet.
Key words: planetary systems – planetary systems: formation – stars: individual (HAT-P-7) – stars: rotation
Online-only material: machine-readable table

1. INTRODUCTION vectors is likely greater than 86◦ , indicating that the orbit is
either retrograde (ψ > 90◦ ) or nearly polar (ψ ≈ 90◦ ). We also
In the solar system, the planetary orbits are well aligned and present evidence for a third body in the system, which may be an
prograde, revolving in the same direction as the rotation of the additional planet or a companion star. We present spectroscopic
Sun. This fact inspired the “nebular hypothesis” that the Sun and data in Section 2, photometric data in Section 3, a joint analysis
planets formed from a single spinning disk (Laplace 1796). One of both types of data in Section 4, and a discussion of the results
might also expect exoplanetary orbits to be well aligned with in Section 5.
their parent stars, and indeed this is true of most systems for
which it has been possible to compare the directions of orbital 2. RADIAL VELOCITIES
motion and stellar rotation (Fabrycky & Winn 2009; Le Bouquin
We observed HAT-P-7 with the High Resolution Spectrograph
et al. 2009). However, there are at least three exoplanets for
(HIRES) on the Keck I 10 m telescope, and the High Dispersion
which the orbit is tilted by a larger angle than any of the planets
Spectrograph (HDS) on the Subaru 8 m telescope. The planet’s
in the solar system: XO-3b (Hébrard et al. 2008; Winn et al.
discoverers (Pál et al. 2008; hereafter P08) obtained eight
2009a), HD 80606b (Moutou et al. 2009; Pont et al. 2009; Winn
HIRES spectra in 2007, to which we add nine spectra from
et al. 2009b), and WASP-14b (Johnson et al. 2009).
2009. All but one of the HIRES spectra were acquired outside
Still, all of those systems are consistent with prograde orbits,
of transits. Of the 49 HDS spectra, 9 were obtained on 2009
with the largest minimum angle between the stellar-rotational
June 17 and 40 were obtained on 2009 July 1. The second of
and orbital angular momentum vectors of about 37◦ , for XO-3b
these nights spanned a transit.
(Winn et al. 2009a). The reason why only the minimum angle
The instrument settings and observing procedures in both
is known is that the evidence for misalignment is based on the
2007 and 2009 were identical to those used by the California
eponymous effect of Rossiter (1924) and McLaughlin (1924), an
Planet Search (CPS; Howard et al. 2009). We placed an
anomalous Doppler shift observed during planetary transits that
iodine gas absorption cell into the optical path, to calibrate the
is sensitive only to the angle between the sky projections of the
instrumental response and wavelength scale. The radial velocity
two vectors. The true spin–orbit angle may be larger, depending
(RV) of each spectrum was measured with respect to an iodine-
on the unknown inclination angle of the stellar rotation axis with
free template spectrum, using the algorithm of Butler et al.
respect to the line of sight.
(2006) with subsequent improvements. Measurement errors
In this Letter, we present evidence of a very large spin–orbit
were estimated from the scatter in the fits to individual spectral
misalignment for HAT-P-7b, a planet of mass 1.8 MJup and radius
segments spanning a few angstroms. The RVs are given in
1.4 RJup in a 2.2-day orbit around an F6V star with mass 1.5 M
Table 1.
and radius 1.8 R (Pál et al. 2008). We find the angle between
the sky-projected angular momentum vectors to be 182.◦ 5 ± 9.◦ 4. 2.1. Evidence for a Third Body
Furthermore, we show that the true angle ψ between those
Figure 1 shows the RVs over the two-year span of the
7 NSF Astronomy and Astrophysics Postdoctoral Fellow. observations. Figure 2 shows the RVs as a function of orbital
8 Townes Postdoctoral Fellow. phase, fitted with two different models. The first model is a single

L99
L100 WINN ET AL. Vol. 703

Figure 1. Long-term radial velocity variation of HAT-P-7. (a) Measured RVs.

(b) Residuals (observed − calculated) between the data and the best-fitting
single-planet model. Light blue and dark blue points are HIRES data from 2007
and 2009, respectively.

Table 1
Relative Radial Velocity Measurements of HAT-P-7

HJD RV (m s−1 ) Error (m s−1 ) Spectrographa

2454336.73960 111.08 1.72 1
2454336.85367 58.89 1.78 1
2454337.76212 −236.06 1.70 1
2454338.77440 151.06 1.54 1
2454338.85456 131.12 1.57 1
Figure 2. Phased radial velocity variation of HAT-P-7. (a) Assuming a single
Notes. The RV was measured relative to an arbitrary template spectrum specific Keplerian orbit. (b) Residuals. (c) With an extra parameter γ̇ representing a
to each spectrograph; only the differences among the RVs from a single constant radial acceleration. (d) Residuals. The circles are HIRES data (light
blue from 2007, dark blue from 2009), the green triangles are HDS data from
spectrograph are significant. The uncertainty given in Column 3 is the internal
2009 June 17, and the red squares are HDS data from 2009 July 1.
error only and does not account for any possible “stellar jitter.”
a (1) Keck/HIRES, (2) Subaru/HDS.

(This table is available in its entirety in a machine-readable form in the online
journal. A portion is shown here for guidance regarding its form and content.)
Keplerian orbit, representing the signal of the known planet.
The second model has an additional parameter γ̇ representing
an extra radial acceleration. The second model gives a better
fit to the data, with a root-mean-squared (rms) residual of 7 m
s−1 as compared to 21 m s−1 for the first model. The RVs from
2009 are systematically redshifted by approximately 40 m s−1
compared to RVs from 2007, as evident from the residuals shown
in Figures 1(b) and 2(b). This shift is highly significant, as the
CPS has demonstrated a long-term stability of 2 m s−1 or better
using HIRES and the same reduction codes used here (Howard
et al. 2009).
This RV trend is evidence for an additional companion. Given
the limited time coverage of our observations (two clusters of
points separated by 2 yr), the data are compatible with nearly
any period longer than a few months. A constant acceleration is
the simplest model that fits the excess RV variability, and under
that assumption we may give an order-of-magnitude relation
relating γ̇ to some properties of the companion
Mc sin ic γ̇
∼ = (0.121 ± 0.014) MJup AU−2 , (1)
ac2 G For this purpose, we supplemented the RV data with the most
where Mc is the companion mass, ic its orbital inclination relative
to the line of sight, ac its orbital distance, and the numerical value
is based on our model-fitting results (see Section 4).
2.2. Evidence for a Spin–orbit Misalignment
Figure 3(a) shows the RV data spanning the transit, after
subtracting the orbital RV as computed with the best-fitting
model including γ̇ . We interpret the “anomalous” RV variation
during the transit as the Rossiter–McLaughlin (RM) effect, the
asymmetry in the spectral lines due to the partial eclipse of
the rotating photosphere. In the context of eclipsing binary
stars, the RM effect was predicted by Holt (1893) and observed
definitively by Rossiter (1924) and McLaughlin (1924). For
exoplanets, the RM effect was first observed by Queloz et al.
(2000), and its use in assessing spin–orbit alignment has been
expounded by Ohta et al. (2005) and Gaudi & Winn (2007).
A transiting planet in a well-aligned prograde orbit would first
pass in front of the blueshifted (approaching) half of the star,
causing an anomalous redshift of the observed starlight. Then,
the planet would cross to the redshifted (receding) half of the
star, causing an anomalous blueshift. In contrast, Figure 3(a)
shows a blueshift followed by a redshift: an inversion of the
effect just described. We may conclude, even without any
modeling, that the orbital “north pole” and the stellar “north
pole” point in nearly opposite directions on the sky.
3. PHOTOMETRY
For a quantitative analysis of the RM effect, we wanted to
model both the photometric and spectroscopic transit signals.
precise transit light curve available to us, shown in Figure 3(c).
This light curve is based on observations on UT 2008 September
22 in the Sloan i bandpass, with the Fred L. Whipple 1.2 m
telescope and Keplercam detector, under the auspices of the
Transit Light Curve project (Holman et al. 2006; Winn et al.
2007).
Reduction of the CCD images involved standard procedures
for bias subtraction and flat-field division. Differential aperture
No. 2, 2009 OUTLANDISH ORBIT OF HAT-P-7b L101
Table 2
Model Parameters for HAT-P-7b
Parameter Value
Orbital period, P (d) 2.2047304 ± 0.0000024
Midtransit time (HJD) 2454,731.67929 ± 0.00043
Transit duration (first to fourth contact) (hr) 4.006 ± 0.064
Transit ingress or egress duration (hr) 0.474+0.061
−0.093
Planet-to-star radius ratio, Rp /R 0.0834+0.0012
−0.0021
Orbital inclination, i (deg) 80.8+2.8
−1.2
Scaled semimajor axis, a/R 3.82+0.39
−0.16
Transit impact parameter 0.618+0.039
−0.149
Velocity semiamplitude, K (m s−1 ) 211.8 ± 2.6
Upper limit on eccentricity (99.73% conf.) 0.039
e cos ω −0.0019 ± 0.0077
e sin ω 0.0037 ± 0.0124
Velocity offset, Keck/HIRES (m s−1 ) −51.2 ± 3.6
Velocity offset, Subaru/HDS (m s−1 ) −4.8 ± 2.5
Constant radial acceleration γ̇ (m s−1 yr−1 ) 21.5 ± 2.6
Projected stellar rotation rate, v sin i (km s−1 ) 4.9+1.2
−0.9
Projected spin–orbit angle, λ (deg) 182.5 ± 9.4

parameter. We also included a free parameter for the coefficient

Figure 3. Spectroscopic and photometric transit of HAT-P-7b. (a) The anoma-
lous RV, defined as the output of the Doppler code minus the orbital RV. We
observed a blueshift in the first half of the transit, and a redshift in the second
half of the transit, demonstrating that the sky projections of the orbital and
stellar angular momentum vectors point in opposite directions. (b) Residuals.
Red squares are HDS data from 2009 July 1, and blue circles are HIRES data
obtained on various nights in 2007 and 2009. (c) The relative flux, observed in
the Sloan i band with the FLWO 1.2 m telescope and Keplercam. (d) Residuals.
In panels (a) and (b), the gray line shows the best-fitting model.
photometry was performed for HAT-P-7 and seven comparison
stars. No evidence was found for time-correlated noise using the
“time-averaging” method of Pont et al. (2006), as implemented
by Winn et al. (2009c). The data shown in Figure 3(c) were
corrected for differential extinction as explained in Section 4.
4. JOINT ANALYSIS
We fitted a model to the photometric and RV data in order
to derive quantitative constraints on the angle λ between the
sky projections of the orbital and stellar-rotational angular
momentum vectors. This angle is defined such that λ = 0◦ when
the sky-projected vectors are parallel and λ = 180◦ when they
are antiparallel. Our model for the RM effect was based on the
technique of Winn et al. (2005): we simulated spectra exhibiting
the RM effect at various transit phases, and then measured the
apparent RV of the simulated spectra using the same Doppler
code that is used on actual data. This allowed us to relate the
anomalous RV to the parameters and positions of the star and
planet.
The RV model was the sum of the Keplerian RV and the
anomalous RV due to the RM effect. The photometric model
was based on the analytic equation for the flux of a quadratically
limb-darkened disk with a circular obstruction (Mandel & Agol
2002). As a compromise between fixing the limb-darkening
coefficients u1 and u2 at theoretically calculated values, and
giving them complete freedom, we fixed u1 − u2 at the tabulated
value of 0.3846 (Claret 2004) and allowed u1 + u2 to be a free
of differential air-mass extinction between HAT-P-7 and the
ensemble of comparison stars.
We determined the best values of the model parameters and
their 68.3% confidence limits using a Markov Chain Monte
Carlo algorithm, as described in our previous works (see, e.g.,
Winn et al. 2009a). The likelihood function was given by
exp(−χ 2 /2) with
Nf   Nv  
 fi (obs) − fi (calc) 2  vi (obs) − vi (calc) 2
χ =
2
+ ,
i=1
σf,i i=1
σv,i
(2)
in a self-explanatory notation, with σf,i chosen to be 0.00136,
and σv,i chosen to be the quadrature sum of the RV measurement
error and a “stellar jitter” term of 9.3 m s−1 . These choices led
to χ 2 = Ndof for the minimum-χ 2 model. A Gaussian prior
constraint was imposed upon the orbital period based on the
precise measurement of P08.
Table 2 gives the results for the model parameters. In
particular, the result for λ is 182.◦ 5 ± 9.◦ 4, close to antiparallel,
as anticipated from the qualitative discussion of Section 2.
5. DISCUSSION
Our finding for λ is strongly suggestive of retrograde motion,
in which the orbital motion and stellar rotation are in opposite
directions. However, it must be remembered that λ refers to the
angle between the sky-projected angular momentum vectors.
The true angle ψ between the vectors is given by
cos ψ = cos i cos i + sin i sin i cos λ, (3)
where i and i are the line-of-sight inclinations of the orbital
and stellar angular momentum vectors, respectively. Although i
is known precisely from the transit data, i is unknown.
Supposing i to be drawn from an “isotropic” distribution
(uniform in cos i ), the data demand that ψ > 86.◦ 3 with 99.73%
confidence. Thus, under this assumption, a retrograde orbit is
strongly favored, although a nearly polar and barely prograde
orbit cannot be ruled out.
L102 WINN ET AL.
In fact, there is circumstantial evidence that i is small and
consequently the orbit of HAT-P-7b is nearly polar (ψ ≈ 90◦ ).
The star’s projected rotation rate is unusually low for such a hot
−1
star: v sin i = 4.9+1.2
−0.9 km s in our model, or 3.8 ± 0.5 km s−1
based on the line profile analysis of P08, and Teff = 6350 ± 80 K
according to P08. In the SPOCS catalog of dwarf stars with well-
determined spectroscopic properties (Valenti & Fischer 2005),
only 2 of 37 stars with Teff = 6350 ± 100 K have v sin i <
4.9 km s−1 .
Based on this catalog, the mean rotation rate v for such hot
stars is about 15 km s−1 . As an alternate approach to constraining
ψ, we assumed the rotation velocity v is drawn from a Gaussian
distribution with mean 15 km s−1 and standard deviation
3 km s−1 . The result is ψ = 94.6+5.5
−3.0 deg with 68.3% confidence,
and ψ > 86.◦ 1 with 99.73% confidence. This analysis favors
nearly polar and retrograde orbits. However, one wonders
whether HAT-P-7 should be expected to have a “typical” rotation
rate, given the existence of its short-period planet on a bizarre
orbit. Another caveat is that we found the scaled semimajor axis
a/R to be about 1σ smaller than the finding of P08, suggesting
the star is somewhat larger and more evolved, which would
correspond to a slower expected rotation rate.
Determining i directly may be possible by measuring and
interpreting asteroseismological oscillations (Gizon & Solanki
2003), or photometric modulations produced by starspots (see,
e.g., Henry & Winn 2008). By good fortune, HAT-P-7 is in the
field of view of the Kepler satellite, which is capable of precise
long-term photometry and may be able to accomplish these tasks
(Borucki et al. 2009).
The extraordinary orbit of HAT-P-7b presents an extreme
case for theories of planet formation and subsequent orbital
evolution. HAT-P-7b is a “hot Jupiter” and presumably migrated
inward toward the star after its formation. A prevailing migration
theory involves tidal interactions with the protoplanetary disk,
but such interactions would probably not perturb the initial
coplanarity of the system, and might even bring the system into
closer alignment (Lubow & Ogilvie 2001; Cresswell et al. 2007).
More promising to explain HAT-P-7b are scenarios involving
few-body dynamics, as those scenarios are expected to produce
misalignments. In one scenario, close encounters between
planets throw a planet inward, where its orbit is ultimately
shrunk and circularized by tidal dissipation (Chatterjee et al.
2008; Jurić & Tremaine 2008). Another idea is based on the
Kozai (1962) effect, whereby the gravitational force from a
distant body on a highly inclined orbit strongly modulates an
inner planet’s orbital eccentricity and inclination (Fabrycky &
Tremaine 2007). Recent calculations showed that a combination
of planet–planet scattering, the Kozai effect, and tidal friction
can lead to nearly circular retrograde orbits (Nagasawa et al.
2008). A third proposed scenario involves an inward-migrating
outer planet that captures an inner planet into a mean motion
resonance; if the inner planet avoids being ejected or consumed
by the star, it may be released on a nearly circular retrograde
orbit (Yu & Tremaine 2001).
The prospect of explaining HAT-P-7b’s orbit through few-
body dynamics lends extra importance to measuring the mass
and orbital parameters of the third body. If it turns out to be
a planet, then HAT-P-7b will be only the second known case
of a transiting planet accompanied by another planet, the first
being HAT-P-13b (Bakos et al. 2009). Such systems are highly
desirable because the unusually precise measurements enabled
by transit observations can be used to determine whether the
orbits are coplanar and give clues about the system’s dynamical
history (Fabrycky 2009).
Vol. 703
We are grateful to Yasushi Suto and Ed Turner for stimulating
our interest in this subject; Norio Narita and his team for sharing
their data in advance of publication; Dan Fabrycky, András
Pál, Darin Raggozine, Scott Tremaine, Bill Welsh, and the
anonymous referee for helpful comments on the manuscript;
Akito Tajitsu, Tae-Soo Pyo, Mark Everett, Howard Isaacson,
and Zach Gazak for assistance with observing; Gáspár Bakos
and Joel Hartman for help obtaining telescope time; Eric Gaidos
and Debra Fischer for trading telescope time on short notice; and
Hector Balbontin for hospitality at Las Campanas Observatory
where this manuscript was written.
Some of the data presented herein were obtained at the W. M.
Keck Observatory, which is operated as a scientific partnership
among the California Institute of Technology, the University of
California, and the National Aeronautics and Space Administra-
tion, and was made possible by the generous financial support of
the W. M. Keck Foundation. We extend special thanks to those
of Hawaiian ancestry on whose sacred mountain of Mauna Kea
we are privileged to be guests. Without their generous hospi-
tality, the Keck observations presented herein would not have
been possible. J.A.J. gratefully acknowledges support from the
NSF Astronomy and Astrophysics Postdoctoral Fellowship pro-
gram (grant AST-0702821). S.A. acknowledges the support of
the Netherlands Organisation for Scientific Research (NWO).
J.N.W. gratefully acknowledges support from the NASA Origins
program through awards NNX09AD36G and NNX09AB33G,
and from an MIT Class of 1942 Career Development Professor-
ship.
Facilities: Subaru (HDS), Keck I (HIRES), FLWO 1.2 m
(Keplercam)
Note added after submission. Narita et al. (2009) report in-
dependent evidence for a retrograde or polar orbit of HAT-P-
7b, based on Subaru/HDS spectra spanning the transit of 2008
May 30.
REFERENCES
Bakos, G. A., et al. 2009, arXiv:0907.3525
Borucki, W. J., et al. 2009, Science, 325, 709
Butler, R. P., et al. 2006, ApJ, 646, 505
Chatterjee, S., Ford, E. B., Matsumura, S., & Rasio, F. A. 2008, ApJ, 686,
580
Claret, A. 2004, A&A, 428, 1001
Cresswell, P., Dirksen, G., Kley, W., & Nelson, R. P. 2007, A&A, 473, 329
Fabrycky, D. C. 2009, in IAU Symp. 253, ed. F. Pont, D. D. Sasselov, & M. J.
Holman (Cambridge: Cambridge Univ. Press), 173
Fabrycky, D., & Tremaine, S. 2007, ApJ, 669, 1298
Fabrycky, D. C., & Winn, J. N. 2009, ApJ, 696, 1230
Gaudi, B. S., & Winn, J. N. 2007, ApJ, 655, 550
Gizon, L., & Solanki, S. K. 2003, ApJ, 589, 1009
Hébrard, G., et al. 2008, A&A, 488, 763
Henry, G. W., & Winn, J. N. 2008, AJ, 135, 68
Holman, M. J., et al. 2006, ApJ, 652, 1715
Holt, J. R. 1893, A&A, 12, 646
Howard, A. W., et al. 2009, ApJ, 696, 75
Johnson, J. A., Winn, J. N., Albrecht, S., Howard, A. W., Marcy, G. W., &
Gazak, J. Z. 2009, PASP, in press (arXiv:0907.5204)
Jurić, M., & Tremaine, S. 2008, ApJ, 686, 603
Kozai, Y. 1962, AJ, 67, 591
Laplace, P. S. 1796, Exposition du Système du Monde (Paris: Cercle-Social)
Le Bouquin, J.-B., Absil, O., Benisty, M., Massi, F., Mérand, A., & Stefl, S.
2009, A&A, 498, L41
Lubow, S. H., & Ogilvie, G. I. 2001, ApJ, 560, 997
Mandel, K., & Agol, E. 2002, ApJ, 580, L171
McLaughlin, D. B. 1924, ApJ, 60, 22
Moutou, C., et al. 2009, A&A, 498, L5
Nagasawa, M., Ida, S., & Bessho, T. 2008, ApJ, 678, 498
No. 2, 2009 OUTLANDISH ORBIT OF HAT-P-7b L103
Narita, N., Sato, B., Hirano, T., & Tamura, M. 2009, PASJ, in press Rossiter, R. A. 1924, ApJ, 60, 15
(arXiv:0908.1673) Valenti, J. A., & Fischer, D. A. 2005, ApJS, 159, 141
Ohta, Y., Taruya, A., & Suto, Y. 2005, ApJ, 622, 1118 Winn, J. N., Holman, M. J., & Fuentes, C. I. 2007, AJ, 133, 11
Pál, A., et al. 2008, ApJ, 680, 1450 Winn, J. N., et al. 2005, ApJ, 631, 1215
Pont, F., Zucker, S., & Queloz, D. 2006, MNRAS, 373, 231 Winn, J. N., et al. 2009a, ApJ, 700, 302
Pont, F., et al. 2009, A&A, 502, 695 Winn, J. N., et al. 2009b, ApJ, in press (arXiv:0907.5205)
Queloz, D., Eggenberger, A., Mayor, M., Perrier, C., Beuzit, J. L., Naef, D., Winn, J. N., et al. 2009c, ApJ, 693, 794
Sivan, J. P., & Udry, S. 2000, A&A, 359, L13 Yu, Q., & Tremaine, S. 2001, AJ, 121, 1736